Occupant observation device

1. An occupant observation device comprising: an imager configured to capture an image of a head of an occupant of a vehicle; and a processor configured to: detect at least a part of eyes of the occupant in the image captured by the imager, based on a distribution of edges extracted from the image, an edge of the distribution of edges being a pixel or a pixel group in which a difference in a pixel value between the pixel or the pixel group and surrounding pixel values of surrounding pixels is greater than a reference value; detect at least a part of a mouth of the occupant in the image captured by the imager; and estimate a condition of the occupant based on a result of the eye detection and a result of the mouth detection, and output a result of estimation to an in-vehicle device, wherein the processor changes a ratio of reflecting the result of the eye detection and a ratio of reflecting the result of the mouth detection based on the estimate of the condition of the occupant, wherein the processor reduces the ratio of reflecting the result of the eye detection and increases the ratio of reflecting the result to the mouth detection in the estimation of the condition of the occupant, based on the result of the eye detection or a result of a process performed based on the result of the eye detection.

2. The occupant observation device according to claim 1 , wherein the processor detects at least a part of a contour of the eyes of the occupant, derives an eye opening rate of the eyes of the occupant, based on a positional relationship of a plurality of feature points in the detected contour, and estimates the condition of the occupant based on the eye opening rate of the eyes of the occupant and the result of the mouth detection, and the result of the process performed based on the result of the eye detection is the eye opening rate of the eyes of the occupant.

3. The occupant observation device according to claim 2 , wherein when a condition in which the eye opening rate of the eyes of the occupant is equal to or less than a predetermined degree continues for a predetermined time or more, the processor changes the ratio of reflecting the result of the eye detection and the ratio of the result of the mouth detection in the estimation of the condition of the occupant.

4. The occupant observation device according to claim 2 , wherein the condition of the occupant is a drowsiness of the occupant, and the processor digitizes the drowsiness of the occupant, by calculating a weighted sum of a reciprocal of the eye opening rate and the number of yawns in an observation period.

5. The occupant observation device according to claim 1 , wherein the processor reduces the ratio of reflecting the result of the eye detection, when a pupil cannot be distinguished or a white of the eye and an iris cannot be distinguished.

6. The occupant observation device according to claim 1 , wherein the processor extracts edges using an edge extraction filter and detects at least a part of the eyes of the occupant in the image on the basis of the distribution of edges, and reduces the ratio of reflecting the result of the eye detection, when a fitting rate of the edges to the contour of the eye is lower than a reference, or a variation of the distribution of the edges to the fitting curve is larger than a reference.

7. An occupant observation device comprising: an imager configured to capture an image of a head of an occupant of a vehicle; and a processor configured to: detect, as a result of an eye detection, at least a part of eyes of the occupant in an image captured by the imager; detect, as a result of a mouth detection, at least a part of the mouth of the occupant in the image captured by the imager; and estimate a condition of the occupant based the result of the eye detection and the result of the mouth detection, wherein the processor changes a ratio of reflecting each of the result of the eye detection and the result of the mouth detection in an estimation of the condition of the occupant, based on the result of the eye detection or a result of a process performed based on the result of the eye detection, and wherein the processor reduces the ratio of reflecting the result of the eye detection, when a pupil cannot be distinguished or a white of the eye and an iris cannot be distinguished.

8. An occupant observation device comprising: an imager configured to capture an image of a head of an occupant of a vehicle; and a processor configured to: detect, as a result of a eye detection, at least a part of eyes of the occupant in an image captured by the imager; detect, as a result of a mouth detection, at least a part of the mouth of the occupant in the image captured by the imager; and estimate a condition of the occupant based on the result of the eye detection and the result of the mouth detection, wherein the processor changes a ratio of reflecting each of the result of the eye detection and the result of the mouth detection in an estimation of the condition of the occupant, based on the result of the eye detection or a result of a process performed based on the detection result of the eye detection; and wherein the processor extracts edges using an edge extraction filter and detects at least a part of the eyes of the occupant in the image based on a distribution of edges, and reduces the ratio of reflecting the result of the eye detection, when a fitting rate of the edges to the contour of the eye is lower than a reference, or a variation of the distribution of the edges to the fitting curve is larger than the reference.